(Last updated on 2020-04-26)
This replicates comparison of the epidemic curve among South Korea, Singapore, Taiwan, and OECD country comparison analysis, using US state-level data.
Hover over each figure to see values and more options. States are color-coded by USDA’s geogrpahic region.
See data sources and methods at the end.
Cumulative incidence rate: number of confirmed cases per 100,000 population. It may not be a good indicator to compare, considering vastly different testing rates - especially in the beginning of the epidemic. Also, definitions of confirmed cases can be slightly different across states.
Mortality rate: number of COVID-19 deaths per 100,000 population. Considering different testing rates and this may be the most appropriate indicator to compare impact of the epidemic across states.
And, for those who wonder about Case fatality rate (CFR): number of deaths per 100 confirmed cases. CFR is sensitive to demographic and clinical characteristics of patients. CFR can be high in states where testing has been limited to only people with clinical symptoms.
The epidemic curve in this analysis was constructed based on the number of new confirmed cases each day. Considering vastly different population sizes, the number of new confirmed cases each day per 100,000 population was used, instead of the absolute number. In addition, 7-day rolling averages were used to avoid any isolated peaks/drops, which can be caused by various reasons other than the true course of the epidemic itself (e.g., changes in definition, and lab process delay).
Let’s focus on the first wave of the epidemic, which we define to start when cumulative incidence exceeds 1 confirmed case per 100,000 population. The peak is when the number of new confirmed cases is at its maximum. There are three distinctive phases:
With successful control of the epidemic, we will see both narrow and also short peak.
The data (i.e., smoothed number of new confirmed cases each day per 100,000 population) show that Vermont has well passed the peak, and is approaching the stable phase as long as the pace continues. The cumulative incidence rate exceeded 1 per 100,000 population on 2020-03-16, and the daily new cases reached the peak (6.6 new confirmed cases per 100,000 population) on 2020-04-07.
Holding beginning of the first wave on day 0, the following figure compares the length and height of the peak across the states.
Two states, Montana and Hawaii, have entered the stable phase.
However, some would reach the stable phase soon if the pace continues (e.g., Vermont).
Still, most states passed the peak only recently. In these states, the peak can move up since fluctuating increases were observed in other states/countries.
Some have entered the first wave, and the curve is going up slowly.
States are presented in groups, due to varying range of peak height. Top panel has states with relatively lower peak, and the bottom panel shows states with higher peak.
To summarize, below first shows the timeline of the first wave and its phases.
* The peak length is still to be determined in many states (with no yellow dot below).
The following compares the height of the peak (the maximum number of new confirmed cases per 100,000 population) by country.
And, finally, if we put together the height of the peak (on a log scale, Y axis) and time to the peak (distance between the red and orange dots above)…
METHODS
Data:
1. All COVID-19 data (i.e., cumulative confirmed cases and deaths by day) come from JHU/CSSE. Accessed on 2020-04-26.
2. All data on country population come from UN World Population Prospects 2019 Revision. Accessed on April 18, 2020.
Measures:
The number of new confirmed cases on each date was calculated based on the difference between cumulative numbers over two days. Then, a 7-day rolling average was calculated for, hereinafter referred to as the smoothed number of new confirmed cases. Then, the smoothed number was divided by the total population in the country, as the smoothed number of new confiremd cases per 100,000 popualtion.
* The first wave was defined to start when the cumulative incidence rate exceeds 1 per 100,000 population. The smoothed number of new confirmed cases at the start of the first wave varies by country, but it averages around 0.5 per 100,000.
* A peak date of the wave was the date when the smoothed number of new confiremd cases per 100,000 popualtion was the maximum.
* A date entering the stable phase was when the smoothed number of new confiremd cases per 100,000 popualtion was less than the number on the start date of the wave.
COVID-mortality (either CFR or preferably COVID-specific mortality rate in the population) may be more comparable indicator, given considerably different testing strategies and testing rates. However, states are currently at different stages of the curve, and comparison of mortality data would be more appropriate once most states are in a similar phase of the epidemic (i.e., well in to the stable phase).
See GitHub for data, code, and more information. For typos, errors, and questions, contact me at www.isquared.global
Making Data Delicious, One Byte at a Time, in good times and bad times.